How Much Will AI Affect Critical Thinking?

At the risk of seeming flippant, artificial intelligence probably will not negatively affect “critical thinking” very much. 

The reason has less to do with “AI giving one the answers” and more the reality that critical thinking skills likely will remain a bell curve. Though controversial for social reasons, bell curves are widely found in nature.  

source 

Also, such skills are domain specific, researchers suggest. People reason using their knowledge of a particular subject matter, and skills learned in one domain do not automatically transfer to another, researchers suggest. 

not a capability that “most students” automatically acquire to a high level simply by going through school or college. Programs in school meant to teach critical thinking skills have had limited success, some suggest. 

Richard Arum and Josipa Roksa's longitudinal Collegiate Learning Assessment project, which followed more than 2,300 students at 24 four-year institutions, found that 45 percent of students showed no statistically significant improvement in critical thinking, complex reasoning and written communication during their first two years of college.

And the pattern of skill improvements might well suggest that preexisting conditions (skills, aptitude, experience) shape the extent of skills acquisition. 

Another study found some changes after four years of college work, though generally of “small” dimensions. The average amount of improvement is modest, and the distribution of outcomes is wide, which is about what a bell curve distribution might suggest would be the case. 

In other words, the causality might be that low critical thinking skills lead to high AI dependence, rather than high AI dependence leading to lower critical thinking.

Student type	Pre-AI ability	Likely AI-era behavior
Low reasoning	Low	Accepts plausible answers with little checking
Lower-middle	Moderate	Uses AI effectively for routine tasks but misses subtle errors
Middle	Moderate	Can check obvious errors, struggles with sophisticated ones
Upper-middle	High	Uses AI iteratively and verifies important claims
High	Very high	Challenges AI, generates counterarguments, identifies subtle errors
Expert	Very high + domain knowledge	Uses AI as an intellectual multiplier

A 2024 study suggests some over-reliance and reduced independent decision-making, critical thinking and analytical reasoning among students using AI. Lower cognitive proficiency was found in another study.  

But there probably is not a single bell curve of "critical thinking." There could be separate distributions for different capabilities such as reasoning, knowledge, metacognition, skepticism, evidence evaluation and domain expertise.

Critical thinking might involve some combination of:

• identifying assumptions

• evaluating evidence

• distinguishing correlation from causation

• recognizing alternative explanations

• detecting logical errors

• comparing competing interpretations

• asking appropriate questions

• judging the reliability of information

• drawing justified conclusions

• revising one's beliefs when evidence changes.

But most of us do not use such skills, most of the time, for most daily interactions. In fact, such reasoning might be fairly rare, in real life, for most people, most of the time. 

The unsettled issue is whether, and to what extent, AI will change the shape of the cognitive skills curve. In principle, AI could flatten the curve, if it raises all learner capabilities. But it might also accentuate differences if AI is differentially useful. In other words, the curve might be reshaped.

But such a distribution might also be unusual in nature. If preexisting skills also tend to predict post-AI skill levels, then a standard bell distribution might be the result. 

-----------------